Water flooding injection profile monitoring well point optimization method

Abstract

The invention relates to the technical field of oilfield development, in particular to a water flooding injection profile monitoring well point optimization method. The method comprises the following steps: acquiring feature attribute data of an oil-water well in a to-be-analyzed area, and screening the feature attribute data to obtain screened feature attribute data; performing principal component analysis on the screened characteristic attribute data, and determining principal component data of each well according to an analysis result; according to the difference of the corresponding principal component data between different oil-water wells, classification under different monitoring well numbers is carried out, and monitoring well distribution results under different monitoring well numbers in the to-be-analyzed area are determined according to classification results; monitoring errors of the monitoring network under different monitoring well numbers are analyzed according to the monitoring well distribution result, the optimal monitoring well number is determined according to the analysis result, and monitoring well point deployment is conducted through the optimal monitoring well number. According to the invention, the water-drive injection profile monitoring precision of the oil field is improved.

Description

Technical Field

[0001] This invention relates to the field of oilfield development technology, and specifically to a method for optimizing well points in water drive injection profile monitoring. Background Technology

[0002] Injection profile testing data is crucial for analyzing reservoir water washing and production status in oilfields. For high water-cut oilfields in the development stage, this data is of paramount importance for stable production and precise control of the development side. Currently, the injection profile monitoring rate is generally around 40% in actual development. However, with the continuous increase in the number of development wells and frequent adjustments to well network densification, multiple overlapping well networks lead to complex underground oil and water distribution. Significant changes have occurred in the injection-production system structure and reservoir flow characteristics, increasing the actual monitoring requirements. This has highlighted the growing contradiction between cost control and monitoring needs. Therefore, achieving comprehensive oilfield development process requirements—including development situation analysis, target setting, well and layer selection for potential tapping measures, and effect evaluation—within the constraints of limited well monitoring has become a key challenge in the field of oilfield dynamic monitoring.

[0003] A review of relevant domestic and international research reveals a lack of effective optimization methods for oilfield monitoring systems. Existing research typically focuses on innovative pressure monitoring methods, such as the uniform grid thinning method and the minimum variance optimization method. These methods primarily involve iterative calculations, comparing the pressure of optimized wells with the average pressure across the entire area, and determining the final monitoring ratio based on the comparison results. According to the "Oilfield Development Management Outline" of CNPC, for medium- and high-permeability sandstone and conglomerate reservoirs, wells with more than 50% of their injection wells in operation should have their injection profiles monitored annually. For wells with severe sand production and for heavy oil reservoirs developed using conventional methods, wells with more than 30% of their injection wells in operation should also have their injection profiles monitored annually. For complex fault-block reservoirs, the monitoring ratio can be appropriately reduced. Currently, the oilfield development has entered the late stage of ultra-high water cut. The current monitoring scheme cannot effectively meet the needs of dynamic status analysis, fine potential tapping, water control and efficiency improvement under the joint development of multiple layers, and can also maximize the utilization of monitoring data. Therefore, it is difficult to obtain an effective monitoring well ratio and monitoring well point distribution, resulting in low monitoring accuracy of the water drive injection profile of the oilfield. Summary of the Invention

[0004] To address the technical problem of low accuracy in monitoring water-drive injection profiles in oilfields due to the inability of traditional water-drive injection profile monitoring techniques to accurately determine the optimal ratio and distribution of monitoring wells, the present invention aims to provide a method for optimizing water-drive injection profile monitoring wells. The specific technical solution adopted is as follows:

[0005] This invention proposes a method for optimizing wellpoints in water-drive injection profile monitoring, the method comprising:

[0006] Obtain characteristic attribute data of oil and water wells in the area to be analyzed, and filter the characteristic attribute data according to the correlation differences between the characteristic attribute data to obtain the filtered characteristic attribute data;

[0007] Principal component analysis was performed on the filtered feature attribute data, and the principal components were filtered based on the analysis results to determine the principal component data for each well.

[0008] Based on the differences in principal component data between different oil and water wells, classification is performed under different numbers of monitoring wells. The distribution of monitoring wells under different numbers of monitoring wells in the area to be analyzed is determined by the classification results.

[0009] Based on the distribution results of the monitoring wells, the monitoring error of the monitoring network under different numbers of monitoring wells is analyzed. Based on the analysis results, the optimal number of monitoring wells is determined, and the monitoring wells are deployed using the optimal number of monitoring wells.

[0010] Furthermore, the characteristic attribute data includes at least: sandstone thickness, effective sandstone thickness of sedimentary facies A, effective thickness, effective sandstone thickness of sedimentary facies B, effective thickness of more than 1 meter, effective sandstone thickness of sedimentary facies C, effective thickness of 0.5-1 meter, number of perforated layers, effective thickness of 0-0.5 meter, maximum perforated stratigraphic coefficient, thickness outside the surface, coefficient of variation of stratigraphic coefficient, effective thickness with permeability of more than 0.3, monthly average daily injection volume, effective thickness with permeability of 0.1-0.3, monthly average oil pressure, effective thickness with permeability of 0-0.1, and stratigraphic coefficient.

[0011] Furthermore, the process of filtering the feature attribute data includes:

[0012] Take any two correlated feature attribute data as a data group, perform correlation analysis on the two feature attribute data in the data group, and obtain the correlation coefficient of each data group.

[0013] Pre-set a filtering threshold, compare the correlation coefficient with the filtering threshold, and determine the filtered feature attribute data based on the comparison result.

[0014] Furthermore, the specific process of determining the filtered feature attribute data includes:

[0015] Record any one feature attribute data as the target feature attribute data, obtain the correlation coefficient of all data groups corresponding to the target feature attribute data, and remove other feature attribute data from all data groups whose correlation coefficient is greater than the screening threshold, except for the target feature attribute data.

[0016] After elimination, any one of the remaining feature attribute data is recorded as the target feature attribute data. Feature data is eliminated based on the comparison between the correlation coefficient and the screening threshold. This process is repeated until all feature attribute data has been screened.

[0017] Furthermore, the method for determining the principal component data for each well includes:

[0018] Principal component analysis was performed on the filtered feature attribute data to obtain the cumulative variance contribution rate of each feature attribute data, and the principal components were determined based on the cumulative variance contribution rate.

[0019] During the principal component analysis process, the principal component scores corresponding to all principal components of each well are used as the principal component data for each well.

[0020] Furthermore, the method for determining the distribution of monitoring wells under different numbers of monitoring wells within the area to be analyzed includes:

[0021] Pre-set multiple different numbers of monitoring wells;

[0022] Determine the feature difference distance based on the differences in principal component data between different oil and water wells;

[0023] For each number of monitoring wells, all oil and water wells are clustered according to the characteristic difference distance to obtain the monitoring well distribution results for each number of monitoring wells.

[0024] Furthermore, the method for calculating the feature difference distance includes:

[0025] The principal component data of each oil and water well are sorted in descending order of the cumulative variance contribution rate of the principal components, and the resulting vector is called the principal component feature vector.

[0026] Calculate the Euclidean distance between the principal component eigenvectors of different oil and water wells, denoted as the feature difference distance.

[0027] Furthermore, the method for analyzing the monitoring error of the monitoring network under different numbers of monitoring wells includes:

[0028] A well spacing is preset, and a monitoring grid is constructed according to the well spacing for each number of monitoring wells. The nodes in the monitoring grid are nodes to be interpolated.

[0029] Based on the spatial relationship characteristics between oil and water wells in the monitoring grid for each number of monitoring wells, a mutation function is constructed for each principal component data for each number of monitoring wells. The mutation function represents the degree of variation of the principal component data of oil and water wells at different distances.

[0030] The weight parameters for each monitoring well under each principal component analysis were determined based on the variogram function.

[0031] The monitoring error of the monitoring grid for each number of monitoring wells is determined based on the weighting parameters and the variogram.

[0032] Furthermore, the method for constructing the mutation function includes:

[0033] For each type of monitoring network with each number of monitoring wells, the first difference value of each principal component is determined based on the difference between the data of each principal component between any two oil and water wells.

[0034] The variogram value of each principal component at that distance is determined based on the first difference value of each principal component between all two oil and water wells at the same distance;

[0035] Based on the variogram values ​​for all distances corresponding to each principal component, construct the variogram for each number of monitoring wells.

[0036] Furthermore, the process of determining the weight parameters includes:

[0037] A set of Kriging equations is constructed based on the variogram of each principal component data for each number of monitoring wells. The weight parameters of each monitoring well under each principal component analysis are determined by solving the set of Kriging equations for each number of monitoring wells.

[0038] Furthermore, the specific method for determining the monitoring error of the monitoring grid for each number of monitoring wells includes:

[0039] Based on the variogram of each principal component for each number of monitoring wells, obtain the variogram values ​​of each principal component between each interpolation point and each monitoring well;

[0040] Based on all the variogram values ​​of each interpolation point and each monitoring well under each principal component analysis, and the weight parameters of each monitoring well under each principal component analysis, the monitoring error value of each interpolation point under each principal component analysis is obtained;

[0041] Based on the monitoring error values ​​of all interpolation points in the monitoring network for each number of monitoring wells under each principal component analysis, the monitoring error of the monitoring grid for each number of monitoring wells under each principal component analysis is obtained.

[0042] Furthermore, the method for determining the optimal number of monitoring wells includes:

[0043] Based on the number of monitoring wells for each type of monitoring well count and the corresponding monitoring error for each type of principal component analysis, a curve showing the relationship between the number of monitoring wells and the monitoring error of the monitoring grid for each type of principal component analysis is constructed.

[0044] The optimal number of monitoring wells for each principal component analysis is determined based on the aforementioned variation curves.

[0045] The final number of monitoring wells is determined based on the optimal number of monitoring wells corresponding to all principal component analyses.

[0046] The present invention has the following beneficial effects:

[0047] To address the problem of low monitoring accuracy in oilfield water-drive injection profile monitoring due to the inability of traditional water-drive injection profile monitoring technology to accurately determine the optimal proportion and distribution of monitoring wells, this invention proposes a water-drive injection profile monitoring well optimization method. The method involves acquiring characteristic attribute data of oil and water wells within the analysis area, filtering the characteristic attribute data based on the correlation differences between them, and then performing principal component analysis on the filtered data to accurately analyze the differences in characteristic attributes between oil and water wells within the analysis area. Based on the differences in the corresponding principal component data between different oil and water wells, the method classifies the data under different numbers of monitoring wells, determining the distribution of monitoring wells within the analysis area under different numbers of monitoring wells based on the classification results. Finally, the method analyzes the monitoring error of the monitoring network under different numbers of monitoring wells based on the distribution results, thereby constructing an evaluation method for the optimal monitoring network quality. This method accurately analyzes the monitoring error characteristics of the analysis area, determines the number of monitoring wells for the injection profile, and accurately obtains the optimal monitoring wells that reflect the actual situation. Attached Figure Description

[0048] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a flowchart of a water-drive injection profile monitoring wellpoint optimization method provided in one embodiment of the present invention;

[0050] Figure 2 This is a flowchart illustrating a method for analyzing the monitoring error of a monitoring network with different numbers of monitoring wells, provided in one embodiment of the present invention.

[0051] Figure 3 This is a schematic diagram illustrating the linear correlation between the effective thickness and the effective sandstone thickness of sedimentary phase B, provided in one embodiment of the present invention.

[0052] Figure 4 This is a schematic diagram illustrating the linear correlation between an effective thickness of more than 1 meter and the effective sandstone thickness of sedimentary phase A, provided in one embodiment of the present invention.

[0053] Figure 5This is a schematic diagram of the distribution of monitoring wells at a 40% ratio, provided as an embodiment of the present invention;

[0054] Figure 6 This is a curve corresponding to the variogram under principal component analysis provided in one embodiment of the present invention;

[0055] Figure 7 This is a comparison chart of the interpolation error result provided in one embodiment of the present invention and the error result after interpolation without adding a fitting variogram.

[0056] Figure 8 This is a schematic diagram of the variation curve between the number of monitoring wells and the monitoring error of the monitoring grid under principal component analysis, provided as an embodiment of the present invention.

[0057] Figure 9 This is a schematic diagram of the rate of change of monitoring error under different numbers of monitoring wells, provided as an embodiment of the present invention. Detailed Implementation

[0058] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a water-drive injection profile monitoring wellpoint optimization method proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0059] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0060] The specific scheme of the water-drive injection profile monitoring well point optimization method provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0061] Please see Figure 1 The diagram illustrates a flowchart of a water-drive injection profile monitoring wellpoint optimization method according to an embodiment of the present invention, the method comprising:

[0062] Step S1: Obtain the characteristic attribute data of oil and water wells in the area to be analyzed, and filter the characteristic attribute data according to the correlation differences between the characteristic attribute data to obtain the filtered characteristic attribute data.

[0063] In order to comprehensively characterize the state of oil and water wells and reflect the characteristics of each well in the interlayer and plane, relevant characteristic attribute data of oil and water wells were extracted by combining relevant domestic and foreign research and oilfield development practices. Furthermore, since there may be redundancy among the extracted characteristic attribute data, the characteristic attribute data were screened to improve the accuracy of subsequent water drive injection profile monitoring well point analysis in the area to be analyzed.

[0064] Preferably, in some possible implementations of the embodiments of the present invention, the characteristic attribute data includes at least: sandstone thickness, effective sandstone thickness of sedimentary facies A, effective thickness, effective sandstone thickness of sedimentary facies B, effective thickness of more than 1 meter, effective sandstone thickness of sedimentary facies C, effective thickness of 0.5-1 meter, number of perforated layers, effective thickness of 0-0.5 meter, maximum perforated formation coefficient, thickness outside the surface, coefficient of variation of formation coefficient, effective thickness with permeability of more than 0.3, monthly average daily injection volume, effective thickness with permeability of 0.1-0.3, monthly average oil pressure, effective thickness with permeability of 0-0.1, and formation coefficient.

[0065] Preferably, in some possible implementations of the embodiments of the present invention, the process of filtering feature attribute data includes:

[0066] First, take any two correlated feature attribute data as a data group, perform correlation analysis on the two feature attribute data in the data group, and obtain the correlation coefficient of each data group.

[0067] Typically, to accurately analyze the characteristics of oil and water wells within the analysis area, a large number of feature attributes are extracted for each well, and some attributes may have information redundancy. Direct analysis would increase computational complexity and interfere with the identification of core features. Therefore, in order to accurately screen key attribute values ​​and improve the efficiency and accuracy of subsequent principal component analysis and cluster analysis, correlation analysis is performed between any two feature attribute values. This is used to accurately judge and analyze redundant information in the feature attribute data, thereby improving the accuracy of subsequent screening of redundant features.

[0068] In a specific implementation of this invention, any two feature attribute data are treated as a data group. The correlation coefficient between different feature attribute data in each data group can be calculated using the Pearson correlation coefficient or the Spearman method. The specific calculation and analysis process is well known to those skilled in the art and will not be described in detail here.

[0069] Next, a filtering threshold is preset, and the correlation coefficient is compared with the filtering threshold. Based on the comparison result, the filtered feature attribute data is determined.

[0070] If the feature attribute data with linear correlation have a significant correlation, then the core information corresponding to the two feature attributes highly overlaps. In this case, one more representative feature attribute can be selected to replace the other feature attribute, that is, the other feature attribute is removed, thereby achieving dimensionality reduction of feature attribute parameters based on correlation assessment. Therefore, a screening threshold is set in advance, and the filtered feature attribute data is determined by comparing the correlation coefficient with the screening threshold, reducing the impact of redundant information on the accuracy of monitoring well point deployment in the area to be analyzed.

[0071] Preferably, in some possible implementations of the embodiments of the present invention, the specific process of determining the filtered feature attribute data includes:

[0072] Record any one of the feature attribute data as the target feature attribute data, obtain the correlation coefficient of all data groups corresponding to the target feature attribute data, and remove the other feature attribute data from all data groups whose correlation coefficient is greater than the screening threshold. After removal, record any one of the remaining feature attribute data as the target feature attribute data, and remove feature data according to the comparison result between the correlation coefficient and the screening threshold. Repeat this process until all feature attribute data has been screened.

[0073] In one specific implementation of this invention, the filtering threshold is set to 0.85. The implementer can adjust the filtering threshold according to the actual requirements for the filtering accuracy of redundant information. If it is necessary to remove the influence of redundant information to a greater extent, a relatively small filtering threshold can be set.

[0074] It should be understood that in the above processing, priority is given to retaining attributes that are more sensitive to the dynamic response of oil and water wells and have clearer geological significance. For example, in the screening of characteristic attribute data, priority is given to retaining core attribute data that are directly related to reservoir utilization and development effects, such as effective thickness, formation coefficient, and monthly average daily injection volume.

[0075] In a specific example, 82 oil and water wells in a certain block of Gaotaizi, Sazhong, were used as the research object. First, relevant characteristic attribute data for each well in the block were extracted. These data included sandstone thickness, effective sandstone thickness of sedimentary facies A, effective thickness, effective sandstone thickness of sedimentary facies B, effective thickness greater than 1 meter, effective sandstone thickness of sedimentary facies C, effective thickness of 0.5-1 meter, number of perforated layers, effective thickness of 0-0.5 meter, maximum perforated formation coefficient, thickness outside the perforation zone, coefficient of variation of formation coefficient, effective thickness with permeability greater than 0.3, average daily injection volume, effective thickness with permeability of 0.1-0.3, average monthly oil pressure, effective thickness with permeability of 0-0.1, and formation coefficient. The linear correlation between different characteristic attributes was analyzed. For example, a schematic diagram of the linear correlation between effective thickness and effective sandstone thickness of sedimentary facies B is shown below. Figure 3As shown, the linear correlation between the effective thickness of more than 1 meter and the effective sandstone thickness of sedimentary facies A is as follows: Figure 4 As shown in Table 1, the feature attributes were filtered through linear correlation analysis. The number of feature attributes was reduced from 20 to 15, thus reducing the impact of redundant information on the calculation and analysis.

[0076] Table 1

[0077]

[0078] Step S2: Perform principal component analysis on the filtered feature attribute data, and filter the principal components based on the analysis results to determine the principal component data for each well.

[0079] After systematically quantifying the linear correlation between different characteristic attributes of oil and water wells and removing redundant information, a set of key characteristic attribute data that can accurately characterize the geological features and production dynamics of oil and water wells was obtained. However, the selected key characteristic attributes still have problems such as high dimensionality and possible weak correlations between some attributes. Furthermore, the influence weight of each attribute on the monitoring target of oil and water wells is difficult to determine accurately. Therefore, principal component analysis was performed on the selected characteristic attribute data. Through analysis, the principal components and corresponding principal component data were determined, which will be used to subsequently analyze the influence of different principal component data on the monitoring accuracy under different numbers of monitoring wells.

[0080] Preferably, in some possible implementations of the embodiments of the present invention, the method for determining the principal component data of each well includes:

[0081] Principal component analysis was performed on the filtered feature attribute data to obtain the cumulative variance contribution rate of each feature attribute data. The principal components were determined based on the cumulative variance contribution rate. The principal component scores corresponding to all principal components of each well during the principal component analysis were taken as the principal component data of each well.

[0082] To accurately select principal component data for key feature attributes, principal component analysis is used for linear transformation and variance decomposition to obtain principal component data that covers core information. This transforms high-dimensional discrete attributes into low-dimensional orthogonal comprehensive analysis, thereby eliminating redundant information and potential interference between attributes, reducing the computational complexity of subsequent cluster analysis and monitoring network error assessment under different numbers of monitoring wells, and improving analysis efficiency. Furthermore, the obtained principal component data can be used to analyze the contribution weight of key attributes to the effective response of oil and water well monitoring errors, highlighting the role of core influencing factors such as formation properties and production dynamics, so that the final selection of monitoring well points can meet the comprehensive needs of oilfield development situation analysis and refined potential tapping.

[0083] In a specific implementation of this invention, all the feature attribute data obtained after screening are used as input. Principal component analysis (PCA) is used to obtain the principal components that are independent of each other and can maximize the carrying of the original information in the high-dimensional key feature attribute set. The principal components are determined based on the criterion that the cumulative variance contribution rate of all principal components is ≥70%, thereby determining the final retained principal components. The principal component scores of each retained principal component are used as the principal component data corresponding to each well. The principal component data can fully cover the core information of the original feature attribute data. The specific process of PCA principal component analysis is well known to those skilled in the art and will not be described in detail here.

[0084] In a specific example, taking 82 oil and water wells in a certain block of Gaotaizi in Sazhong as the research object, principal component analysis was first performed on 15 selected feature attributes. The detailed data obtained after the principal component analysis of one oil and water well is shown in Table 2. According to the criterion that the cumulative variance contribution rate of the principal components is ≥70%, principal component numbers 1-4 were selected as principal components 1-4. The data in the columns corresponding to principal components 1-4 represent the loading coefficients of the feature attribute data corresponding to each principal component number on each selected principal component, reflecting the correlation coefficient between the feature attribute data corresponding to each principal component number and the principal component, and are used to calculate the principal component score of each principal component on all feature attribute data.

[0085] Table 2

[0086]

[0087] Table 3 shows the principal component data of 82 oil and water wells in a certain block of Sazhong Gaotaizi after principal component analysis. Table 2 shows the principal component data of some wells.

[0088] Table 3

[0089]

[0090] Step S3: Classify the monitoring wells based on the differences in principal component data between different oil and water wells under different numbers of monitoring wells, and determine the distribution of monitoring wells under different numbers of monitoring wells in the area to be analyzed based on the classification results.

[0091] During the water drive injection profile monitoring process in the area to be analyzed, if the characteristic attribute data between oil and water wells are similar, that is, the changes in monitoring data on different key characteristic attributes are similar, then one oil or water well can be designated as a monitoring well in the group of oil and water wells with similar characteristic attribute data. Therefore, according to the required number of monitoring wells, oil and water wells can be classified under different numbers of monitoring wells based on the differences in the principal component data corresponding to different oil and water wells. Based on the classification, the distribution results of monitoring wells under different numbers of monitoring wells in the area to be analyzed can be determined, reflecting the optimal distribution results of monitoring wells under each number of monitoring wells requirement.

[0092] Preferably, in some possible implementations of the embodiments of the present invention, the method for determining the distribution results of monitoring wells under different numbers of monitoring wells in the area to be analyzed includes:

[0093] First, a variety of different numbers of monitoring wells are pre-set.

[0094] To determine the optimal distribution of monitoring wells under different numbers of monitoring wells, this invention sets multiple different numbers of monitoring wells, thereby determining the optimal distribution of monitoring wells under each number of monitoring wells.

[0095] In one specific implementation of this invention, starting from 5% of the number of oil and water wells in the area to be analyzed (the rounded result of 5% of the number of wells in the area), the number of wells selected is increased by 5% at a rate of 5% (according to 5%, 10%, ..., 100%) until all oil and water wells are covered, thereby determining the different number of monitoring wells.

[0096] Secondly, the characteristic difference distance is determined based on the differences in principal component data between different oil and water wells.

[0097] Preferably, in some possible implementations of the embodiments of the present invention, the method for calculating the feature difference distance includes:

[0098] The principal component data of each oil and water well are sorted in descending order of the cumulative variance contribution rate of the principal components, and the resulting vector is denoted as the principal component eigenvector. The Euclidean distance between the corresponding principal component eigenvectors of different oil and water wells is calculated and denoted as the eigenvalue distance.

[0099] Principal component data for each oil and water well reflects the core information of the original characteristic attribute data. The smaller the difference in principal component data between different oil and water wells, the smaller the difference in the core information of the original characteristic attribute data between the oil and water wells. Therefore, principal component feature vectors are constructed for the principal component data of each oil and water well according to the variance contribution rate in the principal component analysis process. Then, feature difference distance is obtained based on the difference between the principal component feature vectors, which is used to accurately reflect the difference in the core information of the original characteristic attribute data between different oil and water wells.

[0100] Finally, all oil and water wells were clustered based on the characteristic difference distance for each number of monitoring wells to obtain the monitoring well distribution results for each number of monitoring wells.

[0101] The distribution of monitoring wells in the area to be analyzed varies with the number of monitoring wells, resulting in significant differences in monitoring accuracy. Before determining the optimal number of monitoring wells, it is necessary to determine the distribution of monitoring wells under different numbers of monitoring wells, so as to determine the monitoring error under different monitoring well distributions. Therefore, under different numbers of monitoring wells, oil and water wells are clustered according to the characteristic difference distance between different oil and water wells to obtain the distribution results of monitoring wells under different numbers of monitoring wells.

[0102] In a specific implementation of this invention, for each setting of the number of monitoring wells, the shortest distance method is used to cluster all oil and water wells in the area to be analyzed. The characteristic difference distance between different oil and water wells is used as the distance measurement result between different oil and water wells, and the number of clusters is equal to the number of monitoring wells. After clustering, the monitoring wells are determined by the following selection principle: if a cluster contains 3 or more wells, the well with the smallest Euclidean distance from other wells in the cluster is selected; if a cluster contains only 2 wells, the well with the largest Euclidean distance from other wells in the cluster is selected; if a cluster contains only 1 well, the well is directly listed as a monitoring well.

[0103] In a specific example, 82 oil and water wells in a certain block of Gaotaizi in Sazhong were used as the research object. The distribution of monitored wells in the block under different monitoring ratios was obtained. For example, the schematic diagram of the monitoring well distribution at a 40% ratio is shown below. Figure 5 As shown, Figure 5 This shows the distribution of monitoring wells when the number of monitoring wells is 33.

[0104] Step S4: Analyze the monitoring error of the monitoring network under different numbers of monitoring wells based on the distribution results of the monitoring wells, determine the optimal number of monitoring wells based on the analysis results, and deploy the monitoring wells using the optimal number of monitoring wells.

[0105] In practical applications, the monitoring accuracy of water drive injection profiles varies significantly depending on the number of monitoring wells. In complex fault-block reservoirs, a lower number of monitoring wells is necessary. Furthermore, setting up more monitoring wells increases costs. Therefore, a relatively small number of effective monitoring wells should be set up while meeting monitoring error requirements. To determine the optimal number of monitoring wells for the area to be analyzed, the monitoring error of the monitoring network based on the distribution of monitoring wells with different numbers of wells is analyzed. Based on the analysis results, the optimal number of monitoring wells is determined, thereby optimizing the monitoring well points for the water drive injection profile in the area to be analyzed and enabling efficient monitoring.

[0106] Preferably, in some possible implementations of the embodiments of the present invention, the method for analyzing the monitoring error of the monitoring network under different numbers of monitoring wells is described in [reference needed]. Figure 2 The diagram illustrates a flowchart of a method for analyzing monitoring errors in a monitoring network with different numbers of monitoring wells, according to an embodiment of the present invention. The method includes:

[0107] Step S401: Pre-set the well spacing, construct a monitoring grid for each number of monitoring wells according to the well spacing, and the nodes in the monitoring grid are the nodes to be interpolated.

[0108] The distribution of monitoring wells with different numbers of monitoring wells has a significant impact on the monitoring error of water drive injection profile monitoring. Therefore, in order to evaluate the monitoring error of the monitoring network under each distribution of monitoring wells, a monitoring grid is constructed according to the well spacing of the area to be analyzed for each number of monitoring wells. The nodes in the monitoring grid are interpolation points, which are used to estimate the characteristic attributes of different locations in the area to be analyzed by the monitoring wells with a limited number of monitoring wells. Thus, the monitoring error of the monitoring grid can be evaluated based on the estimation error of the interpolation points with a limited number of monitoring wells, that is, the monitoring error of the area to be analyzed by a limited number of monitoring wells.

[0109] In one specific implementation of this invention, the distance between oil and water wells in the area to be analyzed is calculated, the average of all distance calculation results is used as the well spacing of the monitoring grid, and a square grid is set in the area to be analyzed according to the well spacing. The length of the grid is equal to the numerical value of the well spacing, and the monitoring grid of the entire area to be analyzed is constructed in sequence.

[0110] Step S402: Construct a variogram function for each principal component data under each number of monitoring wells based on the spatial relationship characteristics between oil and water wells in the monitoring grid. The variogram function represents the degree of variation of the principal component data of oil and water wells at different distances.

[0111] To accurately analyze the monitoring error under each number of monitoring wells, it is necessary to perform interpolation estimation on the characteristic attribute data of each interpolation point in the set monitoring grid. The monitoring error of the monitoring grid is then evaluated by the interpolation estimation error. However, in the process of error analysis using interpolation, the interpolation error is related to the number and location of the observation wells, but not to the measured values. Therefore, during the monitoring process, it is necessary to construct a variogram function for each principal component data under each number of monitoring wells based on the spatial relationship characteristics between oil and water wells in the monitoring grid under each number of monitoring wells. Combined with the spatial structure characteristics of the area to be analyzed represented by the variogram function, it is used to analyze the interpolation accuracy under different monitoring grid well number settings, avoiding the waste of costs caused by blindly increasing the number of monitoring wells.

[0112] Preferably, in some possible implementations of the embodiments of the present invention, the method for constructing the mutation function includes:

[0113] For each number of monitoring wells in the monitoring network, the first difference value of each principal component is determined based on the difference between the data of each principal component between any two oil and water wells; the variogram value of each principal component at each distance is determined based on the first difference value of each principal component between all two oil and water wells at the same distance; and the variogram function of each principal component at each number of monitoring wells is constructed based on the variogram values ​​at all distances corresponding to each principal component.

[0114] Because groundwater levels at two different locations in an oil-water well underground system are correlated to some extent; and the principal component data at any location is a random variable, the differences in principal component data of oil-water wells at different distances can be used for quantitative analysis to obtain variogram values, and then variograms can be constructed to describe the structural and random changes of regional variables.

[0115] In a specific implementation of this invention, the relationship of the variogram function for each principal component data is as follows:

[0116] ;in, This indicates that each principal component is at a distance of Variation function values ​​between oil wells; Indicates distance as The total number of oil and water wells; This is the distance between two oil and water wells; and They represent the first The oil well and its distance are Principal component data for each principal component corresponding to oil and water wells. This is the first difference value.

[0117] Step S403: Determine the weight parameters of each monitoring well for each principal component analysis based on the variogram function.

[0118] The variogram function can be used to determine the differences in characteristic attribute information between oil and water wells at different distances within the analysis area. To further determine the importance of different monitoring wells in estimating the characteristic attributes of the interpolation points within the analysis area under principal component analysis, the variogram function is used to evaluate the importance of monitoring wells in interpolation estimation under each principal component analysis, and the weight parameters of monitoring wells under each principal component analysis are obtained.

[0119] Preferably, in some possible implementations of the embodiments of the present invention, the process of determining the weight parameters includes:

[0120] A set of Kriging equations is constructed based on the variogram of each principal component data for each number of monitoring wells. The weight parameters of each monitoring well under each principal component analysis are determined by solving the set of Kriging equations for each number of monitoring wells.

[0121] In one specific implementation of this invention, the Kriging interpolation method can be used to estimate the monitoring grid for each number of monitoring wells. In the process of Kriging interpolation, the construction of the Kriging equation system through the variogram and the corresponding solution process are well known to those skilled in the art and will not be described in detail here.

[0122] Step S404: Determine the monitoring error of the monitoring grid for each number of monitoring wells based on the weighting parameters and the variogram.

[0123] In the process of analyzing the monitoring error of the monitoring grid under each number of monitoring wells, since the interpolation error is related to the number and location of the observation wells and is not related to the measured value, the interpolation error can be calculated and analyzed based on the variogram and the weight parameters of each principal component data under the set number of monitoring wells. This determines the monitoring error of the monitoring grid under each number of monitoring wells and accurately reflects the accuracy of water drive injection profile monitoring in the area to be analyzed under each number of monitoring wells.

[0124] Preferably, in some possible implementations of the embodiments of the present invention, the specific method for determining the monitoring error of the monitoring grid for each number of monitoring wells includes:

[0125] First, based on the variogram of each principal component for each number of monitoring wells, obtain the variogram values ​​of each principal component between each interpolation point and each monitoring well.

[0126] To accurately analyze the accuracy of the monitoring well distribution in estimating the characteristic attributes of the interpolation points under different numbers of monitoring wells, the variogram values ​​of the monitoring wells and the interpolation points on each principal component data are compared. This reflects the estimation error of the interpolation points in the analysis area under the given number of monitoring wells. The larger the variogram value, the greater the deviation in estimating the characteristic attributes of different locations in the analysis area under the current monitoring well distribution.

[0127] In one specific implementation of this invention, the distance between each interpolation point and each monitoring well is calculated, and the distance value is input into the variogram of each principal component to obtain the corresponding variogram value.

[0128] In a specific example, taking 82 oil and water wells in a certain block of Gaotaizi in Sazhong as the research object, the curve corresponding to the variogram under one type of principal component analysis is as follows: Figure 6 As shown.

[0129] Secondly, based on all the variogram values ​​of each interpolation point and each monitoring well under each principal component analysis, and the weight parameters of each monitoring well under each principal component analysis, the monitoring error value of each interpolation point under each principal component analysis is obtained.

[0130] To accurately analyze the monitoring error of the interpolation point under each number of monitoring wells, it is necessary to comprehensively evaluate the monitoring error of each interpolation point under all monitoring well distributions. Based on the variogram values ​​of each interpolation point and each monitoring well under each principal component analysis, as well as the weight parameters of each monitoring well under each principal component analysis, the monitoring error value of each interpolation point under each principal component analysis is obtained, which reflects the error of the monitoring grid under each number of monitoring wells.

[0131] In a specific implementation of this invention, the specific calculation formula for the monitoring error value can be:

[0132] ;in For the first Monitoring error of each interpolation point; It is a Lagrange multiplier; For the first The interpolation point and the first Distance values ​​between monitoring wells; The distance is The The interpolation point and the first The variation function values ​​among the monitoring wells; Indicates the number of monitoring wells. Indicates the first The weight parameters corresponding to each monitoring well.

[0133] It should be understood that the purpose of introducing the Lagrange multiplier in the calculation of the monitoring error value is to ensure that the estimator is unbiased. The Lagrange multiplier can be determined in the process of solving the Kriging equations. The specific determination process is well known to those skilled in the art in solving the Kriging equations and will not be elaborated here.

[0134] Finally, based on the monitoring error values ​​of all interpolation points in the monitoring network for each number of monitoring wells under each principal component analysis, the monitoring error of the monitoring grid for each number of monitoring wells under each principal component analysis is obtained.

[0135] To accurately analyze the global accuracy of the monitoring network and thus optimize the well network density, thereby improving the accuracy of water drive injection profile monitoring in the area to be analyzed, the monitoring error of the monitoring grid under each number of monitoring wells under each principal component analysis is obtained based on the monitoring error values ​​of all interpolation points in the monitoring network under each number of monitoring wells. This is then used for the comprehensive evaluation and analysis of the monitoring error under different numbers of monitoring wells.

[0136] In a specific implementation of this invention, the calculation formula for the monitoring error of the monitoring grid under each principal component analysis for each number of monitoring wells can be:

[0137] ;in, This represents the monitoring error of the monitoring grid for each number of monitoring wells under each principal component analysis. Indicates the first The monitoring error value of each interpolation point; This indicates the total number of interpolation points.

[0138] In a specific example, 82 oil and water wells in a certain block of Gaotaizi, Sazhong, are used as the research object. Based on the distribution of monitoring wells with different numbers of monitoring wells within the block, Kriging interpolation is performed on the entire area using a variogram. The Kriging interpolation error is then evaluated using the variogram. A comparison of the interpolation error results with the error results after interpolation without using a fitted variogram is shown in the figure below. Figure 7 As shown, the left side shows the interpolation results after adding the variogram function, with a maximum interpolation error of 1.8 and an average of 1.05; the right side shows the interpolation results without adding the fitting variogram function, with a maximum interpolation error of 19.1 and an average of 9.2. It can be seen that the error of Kriging interpolation after adding the variogram function is significantly reduced compared to the error of interpolation without adding the fitting variogram function. Therefore, before performing Kriging interpolation on monitoring wells at different proportions across the entire region, it is necessary to calculate the variogram function of each principal component across the entire region, so as to conduct accurate analysis of monitoring errors.

[0139] Preferably, in some possible implementations of the embodiments of the present invention, the method for determining the optimal number of monitoring wells includes:

[0140] First, based on the number of monitoring wells for each type of monitoring well and the corresponding monitoring error for each type of principal component analysis, a curve showing the relationship between the number of monitoring wells and the monitoring error of the monitoring grid under each type of principal component analysis is constructed.

[0141] Since the monitoring accuracy varies greatly depending on the number of monitoring wells during water drive injection profile monitoring, a variation curve is constructed between the number of monitoring wells and the monitoring error of the monitoring grid under each principal component analysis, based on the number of monitoring wells for each type of monitoring well and the corresponding monitoring error under each principal component analysis. The variation curve reflects the correlation and variation characteristics between the number of monitoring wells and the monitoring error, and is used to determine the optimal number of monitoring wells to meet the monitoring error under different principal component analyses.

[0142] In a specific implementation of this invention, 82 oil and water wells in a certain block of Gaotaizi, Sazhong, are taken as the research object. The curve showing the relationship between the number of monitored wells and the monitoring error of the monitoring grid under one type of principal component analysis is as follows: Figure 8 As shown, the curves showing the variation between the number of monitoring wells and the monitoring error of the monitoring grid within the block under principal component analysis can be directly obtained from each curve, which can be used to determine the number of monitoring wells required for monitoring oil and water wells within the block under a given monitoring error.

[0143] Secondly, the optimal number of monitoring wells for each principal component analysis is determined based on the variation curves.

[0144] Typically, the number of monitoring wells corresponding to the inflection point of the change curve can be taken, because from... Figure 8 As can be seen above, when the number of monitoring wells is small at the beginning, the monitoring error decreases rapidly as the number of monitoring wells increases. However, when the number of monitoring wells exceeds a certain threshold, the rate of decrease in monitoring error slows down significantly. In other words, further increasing the number of monitoring wells will not significantly improve the overall monitoring error of the block. Therefore, the optimal number of monitoring wells for each principal component analysis can be determined based on the change curve. That is, the number of monitoring wells corresponding to the inflection point of the change curve is selected as the optimal number of monitoring wells.

[0145] In a specific implementation of this invention, 82 oil and water wells in a certain block of Gaotaizi, Sazhong, are taken as the research object. Figure 8 Taking the first derivative of the change curve, we obtain the curves of the rate of change of monitoring error for different numbers of monitoring wells, as shown in the figure below. Figure 9 As shown.

[0146] Finally, the number of monitoring wells is determined based on the optimal number of monitoring wells corresponding to all principal component analyses.

[0147] In the process of monitoring error analysis of the area to be analyzed, since different principal components reflect different core information features, the monitoring errors under different principal components are analyzed separately. Then, the final number of monitoring wells can be determined by combining the optimal number of monitoring wells determined under different principal component analyses.

[0148] In a specific implementation of this invention, 82 oil and water wells in a certain block of Gaotaizi, Sazhong, are taken as the research object. According to... Figure 9 The point where the rate of change in the curve begins to slow down corresponds to the optimal number of monitoring wells. The specific optimal number of monitoring wells and the corresponding monitoring ratios under different principal components are shown in Table 4. Implementers can determine the optimal number of monitoring wells from the curves of the rate of change of monitoring error under different actual numbers of monitoring wells, based on the monitoring error requirements. If higher accuracy is required, a relatively low threshold for the rate of change of monitoring error can be set. That is, the lower the rate of change of monitoring error, the more monitoring wells are selected, and the higher the accuracy. However, it is necessary to balance the investment cost and monitoring accuracy to avoid blindly increasing the number of monitoring wells.

[0149] Table 4

[0150]

[0151] As shown in Table 4, the optimal number of monitoring wells under different components was determined. The average of the optimal number of monitoring wells under all principal components was taken as the final number of monitoring wells. After calculation and analysis, the final number of monitoring wells determined for 82 oil and water wells in a certain block of Sazhong Gaotaizi was 33, with a monitoring ratio of 39%. After determining the number of monitoring wells, the location of monitoring wells in the block was determined according to the oil layer attribute clustering method. The specific calculation and analysis process of the oil layer attribute clustering method is well known to those skilled in the art, and the detailed process will not be described in detail.

[0152] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0153] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for optimizing wellpoints in water-drive injection profile monitoring, characterized in that, The method includes: Obtain characteristic attribute data of oil and water wells in the area to be analyzed, and filter the characteristic attribute data according to the correlation differences between the characteristic attribute data to obtain the filtered characteristic attribute data; Principal component analysis was performed on the filtered feature attribute data, and the principal components were filtered based on the analysis results to determine the principal component data for each well. Based on the differences in principal component data between different oil and water wells, classification is performed under different numbers of monitoring wells. The distribution of monitoring wells under different numbers of monitoring wells in the area to be analyzed is determined by the classification results. Based on the distribution results of the monitoring wells, the monitoring error of the monitoring network under different numbers of monitoring wells is analyzed. Based on the analysis results, the optimal number of monitoring wells is determined, and the monitoring wells are deployed using the optimal number of monitoring wells.

2. The method for optimizing well points for water-drive injection profile monitoring according to claim 1, characterized in that, The characteristic attribute data includes at least: sandstone thickness, effective sandstone thickness of sedimentary facies A, effective thickness, effective sandstone thickness of sedimentary facies B, effective thickness of more than 1 meter, effective sandstone thickness of sedimentary facies C, effective thickness of 0.5-1 meter, number of perforated layers, effective thickness of 0-0.5 meter, maximum perforated stratigraphic coefficient, thickness outside the surface, coefficient of variation of stratigraphic coefficient, effective thickness with permeability of more than 0.3, monthly average daily injection volume, effective thickness with permeability of 0.1-0.3, monthly average oil pressure, effective thickness with permeability of 0-0.1, and stratigraphic coefficient.

3. The method for optimizing well points for water-drive injection profile monitoring according to claim 1, characterized in that, The process of filtering feature attribute data includes: Take any two correlated feature attribute data as a data group, perform correlation analysis on the two feature attribute data in the data group, and obtain the correlation coefficient of each data group. Pre-set a filtering threshold, compare the correlation coefficient with the filtering threshold, and determine the filtered feature attribute data based on the comparison result.

4. The method for optimizing well points for water-drive injection profile monitoring according to claim 3, characterized in that, The specific process for determining the filtered feature attribute data includes: Record any one feature attribute data as the target feature attribute data, obtain the correlation coefficient of all data groups corresponding to the target feature attribute data, and remove other feature attribute data from all data groups whose correlation coefficient is greater than the screening threshold, except for the target feature attribute data. After elimination, any one of the remaining feature attribute data is recorded as the target feature attribute data. Feature data is eliminated based on the comparison between the correlation coefficient and the screening threshold. This process is repeated until all feature attribute data has been screened.

5. The method for optimizing well points for water-drive injection profile monitoring according to claim 1, characterized in that, The method for determining the principal component data for each well includes: Principal component analysis was performed on the filtered feature attribute data to obtain the cumulative variance contribution rate of each feature attribute data, and the principal components were determined based on the cumulative variance contribution rate. During the principal component analysis process, the principal component scores corresponding to all principal components of each well are used as the principal component data for each well.

6. The method for optimizing well points for water-drive injection profile monitoring according to claim 1, characterized in that, The method for determining the distribution of monitoring wells under different numbers of monitoring wells within the area to be analyzed includes: Pre-set multiple different numbers of monitoring wells; Determine the feature difference distance based on the differences in principal component data between different oil and water wells; For each number of monitoring wells, all oil and water wells are clustered according to the characteristic difference distance to obtain the monitoring well distribution results for each number of monitoring wells.

7. The method for optimizing well points for water-drive injection profile monitoring according to claim 6, characterized in that, The method for calculating the feature difference distance includes: The principal component data of each oil and water well are sorted in descending order of the cumulative variance contribution rate of the principal components, and the resulting vector is called the principal component feature vector. Calculate the Euclidean distance between the principal component eigenvectors of different oil and water wells, denoted as the feature difference distance.

8. The method for optimizing well points for water-drive injection profile monitoring according to claim 1, characterized in that, The method for analyzing the monitoring error of the monitoring network under different numbers of monitoring wells includes: A well spacing is preset, and a monitoring grid is constructed according to the well spacing for each number of monitoring wells. The nodes in the monitoring grid are nodes to be interpolated. Based on the spatial relationship characteristics between oil and water wells in the monitoring grid for each number of monitoring wells, a mutation function is constructed for each principal component data for each number of monitoring wells. The mutation function represents the degree of variation of the principal component data of oil and water wells at different distances. The weight parameters for each monitoring well under each principal component analysis were determined based on the variogram function. The monitoring error of the monitoring grid for each number of monitoring wells is determined based on the weighting parameters and the variogram.

9. The method for optimizing well points for water-drive injection profile monitoring according to claim 8, characterized in that, The method for constructing the mutation function includes: For each type of monitoring network with each number of monitoring wells, the first difference value of each principal component is determined based on the difference between the data of each principal component between any two oil and water wells. The variogram value of each principal component at that distance is determined based on the first difference value of each principal component between all two oil and water wells at the same distance; Based on the variogram values ​​for all distances corresponding to each principal component, construct the variogram for each number of monitoring wells.

10. The method for optimizing well points for water-drive injection profile monitoring according to claim 8, characterized in that, The process of determining the weight parameters includes: A set of Kriging equations is constructed based on the variogram of each principal component data for each number of monitoring wells. The weight parameters of each monitoring well under each principal component analysis are determined by solving the set of Kriging equations for each number of monitoring wells.

11. The method for optimizing well points for water drive injection profile monitoring according to claim 8, characterized in that, The specific methods for determining the monitoring error of the monitoring grid for each number of monitoring wells include: Based on the variogram of each principal component for each number of monitoring wells, obtain the variogram values ​​of each principal component between each interpolation point and each monitoring well; Based on all the variogram values ​​of each interpolation point and each monitoring well under each principal component analysis, and the weight parameters of each monitoring well under each principal component analysis, the monitoring error value of each interpolation point under each principal component analysis is obtained; Based on the monitoring error values ​​of all interpolation points in the monitoring network for each number of monitoring wells under each principal component analysis, the monitoring error of the monitoring grid for each number of monitoring wells under each principal component analysis is obtained.

12. The method for optimizing well points for water drive injection profile monitoring according to claim 1, characterized in that, The method for determining the optimal number of monitoring wells includes: Based on the number of monitoring wells for each type of monitoring well count and the corresponding monitoring error for each type of principal component analysis, a curve showing the relationship between the number of monitoring wells and the monitoring error of the monitoring grid for each type of principal component analysis is constructed. The optimal number of monitoring wells for each principal component analysis is determined based on the aforementioned variation curves. The final number of monitoring wells is determined based on the optimal number of monitoring wells corresponding to all principal component analyses.

Images